[26899] | 1 | import java.io.BufferedReader;
|
---|
| 2 | import java.io.File;
|
---|
| 3 | import java.io.FileOutputStream;
|
---|
| 4 | import java.io.FileWriter;
|
---|
| 5 | import java.io.InputStream;
|
---|
| 6 | import java.io.InputStreamReader;
|
---|
| 7 | import java.io.OutputStreamWriter;
|
---|
| 8 | import java.io.StringWriter;
|
---|
| 9 | import java.net.Authenticator;
|
---|
| 10 | import java.net.HttpURLConnection;
|
---|
| 11 | import java.net.PasswordAuthentication;
|
---|
| 12 | import java.net.URL;
|
---|
| 13 | import java.util.ArrayList;
|
---|
| 14 | import java.util.HashMap;
|
---|
| 15 | import java.util.Iterator;
|
---|
| 16 | import java.util.Map;
|
---|
| 17 |
|
---|
| 18 | import javax.xml.parsers.DocumentBuilder;
|
---|
| 19 | import javax.xml.parsers.DocumentBuilderFactory;
|
---|
| 20 | import javax.xml.parsers.ParserConfigurationException;
|
---|
| 21 |
|
---|
| 22 | //import org.json1.JSONException;
|
---|
| 23 | //import org.json1.JSONTokener;
|
---|
| 24 | import org.w3c.dom.Document;
|
---|
| 25 |
|
---|
| 26 | import com.google.gson.Gson;
|
---|
| 27 | import com.google.gson.JsonArray;
|
---|
| 28 | import com.google.gson.JsonElement;
|
---|
| 29 | import com.google.gson.JsonObject;
|
---|
| 30 | import com.google.gson.JsonParser;
|
---|
| 31 |
|
---|
| 32 |
|
---|
| 33 | public class main
|
---|
| 34 | {
|
---|
| 35 |
|
---|
| 36 | // add your place ids to this
|
---|
| 37 | private static final int[] PLACE_IDS = { 5 };
|
---|
| 38 | private static final boolean DEBUG_MODE = true;
|
---|
| 39 | private static HashMap< String, ArrayList<Location> > place_ids;
|
---|
| 40 |
|
---|
| 41 | public static void main(String[] args)
|
---|
| 42 | {
|
---|
| 43 | setUpProxyConnection();
|
---|
| 44 |
|
---|
| 45 | // key = place_id, value = array list of locations corresponding to that place_id
|
---|
| 46 | place_ids = new HashMap< String, ArrayList<Location> >();
|
---|
| 47 |
|
---|
| 48 | for(int i = 0; i < PLACE_IDS.length; i++)
|
---|
| 49 | {
|
---|
| 50 | // construct the url and get its contents
|
---|
| 51 | String url = "http://thisistheplace.org.nz/index2.php?option=com_kttw&task=display&view=ajaxplacedetails&place_id=" + PLACE_IDS[i] + "&zoom_level=1";
|
---|
| 52 | String url_contents = getFileFromUrl(url);
|
---|
| 53 | url_contents = getJsonFromHtml(url_contents);
|
---|
| 54 | System.out.println(url_contents);
|
---|
| 55 |
|
---|
| 56 | //Gson gson = new Gson();
|
---|
| 57 | //gson.fromjson
|
---|
| 58 |
|
---|
| 59 | JsonElement jElement = new JsonParser().parse(url_contents);
|
---|
| 60 | JsonArray jArray = jElement.getAsJsonArray();
|
---|
| 61 | // the json has 4 top levels, and we want the 4th level
|
---|
| 62 | JsonElement je = jArray.get(3);
|
---|
| 63 |
|
---|
| 64 | // get all the entries
|
---|
| 65 | JsonArray jArray2 = je.getAsJsonArray();
|
---|
| 66 | for(int x = 0; x < jArray2.size(); x++)
|
---|
| 67 | {
|
---|
| 68 |
|
---|
| 69 | // ok we have an entry, now extract information about it
|
---|
| 70 | JsonElement entry = jArray2.get(x);
|
---|
| 71 | JsonObject object_for_entry = entry.getAsJsonObject();
|
---|
| 72 |
|
---|
| 73 | // if the story type is "History" (for historical pieces)
|
---|
| 74 | if( object_for_entry.get("story_type").toString().equals("\"History\"") )
|
---|
| 75 | {
|
---|
| 76 | String name = object_for_entry.get("name").toString();
|
---|
| 77 | String text_content = object_for_entry.get("text_content").toString();
|
---|
| 78 | // now we have to strip the quotes off the data
|
---|
| 79 | name = name.substring(1, name.length()-1);
|
---|
| 80 | text_content = text_content.substring(1, text_content.length()-1);
|
---|
| 81 | //text_content = text_content.replace("\\\"",""");
|
---|
| 82 | //text_content = text_content.replace("<","<");
|
---|
| 83 | //text_content = text_content.replace(">",">");
|
---|
| 84 |
|
---|
| 85 | if(DEBUG_MODE)
|
---|
| 86 | {
|
---|
| 87 | System.out.print(name);
|
---|
| 88 | System.out.println(text_content);
|
---|
| 89 | }
|
---|
| 90 |
|
---|
| 91 | // lets strip off the quotes for place_id
|
---|
| 92 | String place_id = object_for_entry.get("place_id").toString();
|
---|
| 93 | place_id = place_id.substring(1, place_id.length()-1);
|
---|
| 94 |
|
---|
| 95 | // now grab the latlng for this place_id
|
---|
| 96 | url = "http://thisistheplace.org.nz/index2.php?option=com_kttw&task=display&view=ajaxplacedetails&place_id=" + place_id;
|
---|
| 97 | if(DEBUG_MODE)
|
---|
| 98 | System.out.println(url);
|
---|
| 99 | url_contents = getFileFromUrl(url);
|
---|
| 100 | url_contents = getJsonFromHtml(url_contents);
|
---|
| 101 | jElement = new JsonParser().parse(url_contents);
|
---|
| 102 | jArray = jElement.getAsJsonArray();
|
---|
| 103 | je = jArray.get(0);
|
---|
| 104 | object_for_entry = je.getAsJsonObject();
|
---|
| 105 |
|
---|
| 106 | String longitude = object_for_entry.get("longitude").toString();
|
---|
| 107 | String latitude = object_for_entry.get("latitude").toString();
|
---|
| 108 | // again, strip the quotes off the data
|
---|
| 109 | longitude = longitude.substring(1, longitude.length()-1);
|
---|
| 110 | latitude = latitude.substring(1, latitude.length()-1);
|
---|
| 111 |
|
---|
| 112 | if(DEBUG_MODE)
|
---|
| 113 | {
|
---|
| 114 | System.out.println ("For place_id " + place_id + ", we have: " + longitude + ", " + latitude );
|
---|
| 115 | System.out.println("");
|
---|
| 116 | System.out.println("");
|
---|
| 117 | }
|
---|
| 118 |
|
---|
| 119 | // ok, create a location
|
---|
| 120 | Location loc = new Location();
|
---|
| 121 | loc.name = name;
|
---|
| 122 | loc.text_content = text_content;
|
---|
| 123 | loc.latitude = latitude;
|
---|
| 124 | loc.longitude = longitude;
|
---|
| 125 |
|
---|
| 126 | // ok, grab the array list out of our hash map
|
---|
| 127 | ArrayList<Location> list_for_place_id = place_ids.get(place_id);
|
---|
| 128 |
|
---|
| 129 | // if the array list doesn't exist, create it, add our location, and stick the list into the hashmap with the place_id as the key
|
---|
| 130 | if( list_for_place_id == null)
|
---|
| 131 | {
|
---|
| 132 | list_for_place_id = new ArrayList<Location>();
|
---|
| 133 | list_for_place_id.add(loc);
|
---|
| 134 | place_ids.put(place_id, list_for_place_id);
|
---|
| 135 | }
|
---|
| 136 | else
|
---|
| 137 | {
|
---|
| 138 | list_for_place_id.add(loc);
|
---|
| 139 | place_ids.put(place_id, list_for_place_id);
|
---|
| 140 | }
|
---|
| 141 |
|
---|
| 142 | }
|
---|
| 143 | }
|
---|
| 144 |
|
---|
| 145 | }
|
---|
| 146 |
|
---|
| 147 | // now iterate through the hashmap
|
---|
| 148 | processHashMap();
|
---|
| 149 | }
|
---|
| 150 |
|
---|
| 151 | private static void processHashMap()
|
---|
| 152 | {
|
---|
| 153 | File f = new File("articles/");
|
---|
| 154 | if (!f.isDirectory())
|
---|
| 155 | f.mkdir();
|
---|
| 156 |
|
---|
| 157 | Iterator it = place_ids.entrySet().iterator();
|
---|
| 158 | while( it.hasNext() )
|
---|
| 159 | {
|
---|
| 160 | Map.Entry pairs = (Map.Entry) it.next();
|
---|
| 161 |
|
---|
| 162 | String key = (String) pairs.getKey();
|
---|
| 163 | ArrayList<Location> values = (ArrayList<Location>) pairs.getValue();
|
---|
| 164 |
|
---|
| 165 | f = new File("articles/" + key);
|
---|
| 166 | if (!f.isDirectory())
|
---|
| 167 | f.mkdir();
|
---|
| 168 |
|
---|
| 169 | // ok, let's go through all the locations for this particular place_id
|
---|
| 170 | for(int x = 0; x < values.size(); x++)
|
---|
| 171 | {
|
---|
| 172 | Location loc = values.get(x);
|
---|
| 173 |
|
---|
| 174 | writeHtmlForGreenstone(loc, "articles/" + key + "/");
|
---|
| 175 | }
|
---|
| 176 |
|
---|
| 177 |
|
---|
| 178 | }
|
---|
| 179 | }
|
---|
| 180 |
|
---|
| 181 | private static void writeHtmlForGreenstone(Location loc, String dir)
|
---|
| 182 | {
|
---|
| 183 | try
|
---|
| 184 | {
|
---|
| 185 | String name = loc.name;
|
---|
| 186 | String text_content = loc.text_content;
|
---|
| 187 | String longitude = loc.longitude;
|
---|
| 188 | String latitude = loc.latitude;
|
---|
| 189 |
|
---|
| 190 | // Don't want any back-slashed quotes (generated in the JSON syntax) or any HTML elements in the title
|
---|
| 191 | String name_no_formatting = name.replaceAll("<[^>+]>", "");
|
---|
| 192 | name_no_formatting = name_no_formatting.replaceAll("\\\\\"", """);
|
---|
| 193 |
|
---|
| 194 | // For the HTML body, <element> are OK, but still don't want any JSON escaped quotes
|
---|
| 195 | text_content = text_content.replaceAll("\\\\\"", """);
|
---|
| 196 |
|
---|
| 197 | // ok, we'll first write out a basic html file
|
---|
| 198 | FileWriter fw = new FileWriter(dir + loc.name.replace("?", "") + ".html");
|
---|
| 199 | StringWriter sw = new StringWriter();
|
---|
| 200 | sw.write("<html><head><title>"+name_no_formatting+"</title></head><body>");
|
---|
| 201 | sw.write("<h2>"+name+"</h2>");
|
---|
| 202 | sw.write(text_content);
|
---|
| 203 | sw.write("</body></html>");
|
---|
| 204 | fw.write(sw.toString());
|
---|
| 205 | fw.close();
|
---|
| 206 | sw.close();
|
---|
| 207 |
|
---|
| 208 |
|
---|
| 209 | // ok, now lastly we'll write out a metadata.xml
|
---|
| 210 | fw = new FileWriter(dir + "metadata.xml");
|
---|
| 211 | sw = new StringWriter();
|
---|
| 212 |
|
---|
| 213 | sw.write("<?xml version=\"1.0\" encoding=\"UTF-8\"?>");
|
---|
| 214 | sw.write("<!DOCTYPE DirectoryMetadata SYSTEM \"http://greenstone.org/dtd/DirectoryMetadata/1.0/DirectoryMetadata.dtd\">");
|
---|
| 215 | sw.write("<DirectoryMetadata>");
|
---|
| 216 | sw.write("<FileSet>");
|
---|
| 217 | sw.write("<FileName>.*</FileName>");
|
---|
| 218 | sw.write("<Description>");
|
---|
| 219 |
|
---|
| 220 |
|
---|
| 221 | sw.write("<Metadata mode=\"override\" name=\"tp.Longitude\">" + longitude + "</Metadata>");
|
---|
| 222 | sw.write("<Metadata mode=\"override\" name=\"tp.Latitude\">" + latitude + "</Metadata>");
|
---|
| 223 |
|
---|
| 224 | sw.write("</Description>");
|
---|
| 225 | sw.write("</FileSet>");
|
---|
| 226 | sw.write("</DirectoryMetadata>");
|
---|
| 227 |
|
---|
| 228 | fw.write(sw.toString());
|
---|
| 229 | fw.close();
|
---|
| 230 | sw.close();
|
---|
| 231 | }
|
---|
| 232 | catch(Exception ex)
|
---|
| 233 | {
|
---|
| 234 | ex.printStackTrace();
|
---|
| 235 | }
|
---|
| 236 | }
|
---|
| 237 |
|
---|
| 238 | // when thisistheplace.org gives you a json, it's actually a json inside html, so we have to try get the json from the html...
|
---|
| 239 | // (ideally I'd use an html parser but it didn't seem to like the structure of the html...)
|
---|
| 240 | private static String getJsonFromHtml(String url_contents)
|
---|
| 241 | {
|
---|
| 242 | int idx = url_contents.lastIndexOf("<body class=\"contentpane\">");
|
---|
| 243 | int idx2 = url_contents.lastIndexOf("</body>");
|
---|
| 244 | String new_url_contents = url_contents.substring(idx + "<body class=\"contentpane\">".length(), idx2);
|
---|
| 245 |
|
---|
| 246 | return new_url_contents;
|
---|
| 247 | }
|
---|
| 248 |
|
---|
| 249 | private static Document getDocumentFromUrl(String uri)
|
---|
| 250 | {
|
---|
| 251 | Document doc = null;
|
---|
| 252 |
|
---|
| 253 | try
|
---|
| 254 | {
|
---|
| 255 | DocumentBuilderFactory factory = DocumentBuilderFactory.newInstance();
|
---|
| 256 | //factory.setValidating(true);
|
---|
| 257 | factory.setIgnoringElementContentWhitespace(true);
|
---|
| 258 |
|
---|
| 259 | URL url = new URL(uri);
|
---|
| 260 | HttpURLConnection connection = (HttpURLConnection)url.openConnection();
|
---|
| 261 | connection.setRequestMethod("GET");
|
---|
| 262 | connection.connect();
|
---|
| 263 | InputStream stream = connection.getInputStream();
|
---|
| 264 |
|
---|
| 265 | DocumentBuilder builder = factory.newDocumentBuilder();
|
---|
| 266 | doc = builder.parse(stream);
|
---|
| 267 | }
|
---|
| 268 | catch(Exception ex)
|
---|
| 269 | {
|
---|
| 270 | ex.printStackTrace();
|
---|
| 271 | }
|
---|
| 272 |
|
---|
| 273 | return doc;
|
---|
| 274 | }
|
---|
| 275 |
|
---|
| 276 | private static String getFileFromUrl(String uri)
|
---|
| 277 | {
|
---|
| 278 | String totalLine = "";
|
---|
| 279 |
|
---|
| 280 | try
|
---|
| 281 | {
|
---|
| 282 | URL url = new URL(uri);
|
---|
| 283 | BufferedReader in = new BufferedReader(new InputStreamReader(url.openStream()));
|
---|
| 284 |
|
---|
| 285 | String inputLine;
|
---|
| 286 | while ( (inputLine = in.readLine()) != null )
|
---|
| 287 | totalLine += inputLine;
|
---|
| 288 |
|
---|
| 289 | in.close();
|
---|
| 290 | }
|
---|
| 291 | catch(Exception ex)
|
---|
| 292 | {
|
---|
| 293 | ex.printStackTrace();
|
---|
| 294 | }
|
---|
| 295 |
|
---|
| 296 | return totalLine;
|
---|
| 297 | }
|
---|
| 298 |
|
---|
| 299 | private static void setUpProxyConnection()
|
---|
| 300 | {
|
---|
| 301 | System.setProperty("http.proxyHost", "proxy.cms.waikato.ac.nz");
|
---|
| 302 | System.setProperty("http.proxyPort", "3128");
|
---|
| 303 | System.setProperty("http.proxyUser", "cjb60");
|
---|
| 304 | System.setProperty("http.proxyPassword", "cjbeck64oo");
|
---|
| 305 |
|
---|
| 306 | Authenticator.setDefault(
|
---|
| 307 | new Authenticator()
|
---|
| 308 | {
|
---|
| 309 | public PasswordAuthentication getPasswordAuthentication()
|
---|
| 310 | {
|
---|
| 311 | return new PasswordAuthentication("cjb60", "cjbeck64oo".toCharArray());
|
---|
| 312 | }
|
---|
| 313 | }
|
---|
| 314 | );
|
---|
| 315 | }
|
---|
| 316 | }
|
---|